benchgap
Google · model

Gemini 3 Pro benchmark scores

As of 2026-11-21, Gemini 3 Pro (Google) has measured scores on 46 benchmarks and estimated scores on 93 more.

BenchmarkScoreSource
AA-GPQA Diamond90.8%measured
AA-HLE39.7%measured
AA-SciCode47.6%estimated ± 3.8 pp, high confidence
Artificial Analysis Intelligence Index28.0%measured
CritPt9.1%measured
GDPval-AA21.4%estimated ± 6.2 pp, medium confidence
GPQA Diamond89.0%estimated ± 1.4 pp, medium confidence
HLE43.9%estimated ± 6.1 pp, medium confidence
IFBench68.4%estimated ± 6.6 pp, medium confidence
AA-LCR76.0%measured
LiveCodeBench v689.3%estimated ± 2.2 pp, medium confidence
AA-Omniscience Accuracy55.8%measured
SciCode49.5%estimated ± 2.9 pp, medium confidence
BrowseComp64.0%estimated ± 6.2 pp, medium confidence
HLE w/ tools41.3%estimated ± 5.5 pp, medium confidence
IFEval92.8%estimated ± 2.8 pp, high confidence
LongBench v262.5%estimated ± 5.5 pp, medium confidence
τ²-bench results87.1%measured
AA Coding Index61.0%estimated ± 5.6 pp, medium confidence
AA-MMMU-Pro80.2%measured
APEX-Agents-AA15.5%estimated ± 3.6 pp, medium confidence
BioMysteryBench (human-difficult)43.0%estimated ± 6.8 pp, low confidence
DeepSearchQA73.8%estimated ± 8.2 pp, low confidence
SWE-bench Verified81.5%estimated ± 3.2 pp, medium confidence
BFCL v465.1%estimated ± 8.5 pp, low confidence
CyberGym66.1%estimated ± 9.3 pp, medium confidence
JobBench11.4%measured
SWE-bench Pro54.1%estimated ± 2.9 pp, medium confidence
WideResearch72.8%estimated ± 1.2 pp, medium confidence
AA Agentic Index27.1%estimated ± 6.8 pp, medium confidence
MMLU-Pro87.7%estimated ± 4.8 pp, medium confidence
AA-IFBench70.4%measured
FailureSensorIQ69.1%measured
GPQA87.3%estimated ± 1.6 pp, high confidence
MMLU96.3%estimated ± 1.2 pp, medium confidence
Vals GPQA Diamond91.7%measured
Vals LiveCodeBench86.4%measured
Vals MMLU-Pro90.1%measured
MMMLU89.3%estimated ± 0.7 pp, medium confidence
MMMU87.2%measured
Chess Puzzles55.2%estimated ± 12.3 pp, low confidence
AA MMLU-Pro89.8%measured
Gert Labs63.2%measured
AA EnterpriseOps-Gym37.6%estimated ± 1.2 pp, low confidence
ApprenticeBench5.4%estimated ± 3.1 pp, medium confidence
ARC-AGI-175.0%measured
ARC-AGI-231.1%measured
CursorBench 3.150.4%estimated ± 5.8 pp, low confidence
CursorBench 3.256.2%estimated ± 5.4 pp, low confidence
FrontierCode 1.1 Main32.5%estimated ± 5.2 pp, medium confidence
FrontierSWE v28.6%estimated ± 9.5 pp, low confidence
MazeBench0.0%estimated ± 5.1 pp, low confidence
Mystery Game Puzzles16.9%estimated ± 2.2 pp, medium confidence
NYT Connections94.4%measured
OfficeQA Pro57.8%estimated ± 6.8 pp, medium confidence
OSWorld-Verified73.6%estimated ± 5.5 pp, medium confidence
ParseBench-156.0%estimated ± 5.2 pp, low confidence
Vals SWE-bench76.4%measured
VulcanBench v384.6%estimated ± 5.6 pp, low confidence
HealthBench Professional (length-adjusted, Artificial Analysis)54.3%estimated ± 4.3 pp, low confidence
HealthBench Professional (raw, Artificial Analysis)53.4%estimated ± 0.8 pp, low confidence
AA ITBench41.6%estimated ± 2.8 pp, high confidence
Chartography (tools)83.5%estimated ± 4.1 pp, low confidence
CursorBench 4.030.9%estimated ± 3.4 pp, medium confidence
DeepSWE56.3%estimated ± 7.2 pp, low confidence
HLE w/o tools51.1%estimated ± 4.9 pp, high confidence
OSWorld 2.010.7%estimated ± 5.9 pp, low confidence
PostTrainBench v1.115.0%estimated ± 1.3 pp, low confidence
SpireClimb5.8%estimated ± 7.2 pp, medium confidence
StealthMark58.3%estimated ± 10.7 pp, low confidence
GDP.pdf16.4%estimated ± 4.1 pp, low confidence
ARC-AGI-37.6%estimated ± 8.6 pp, low confidence
ARC-AGI-3 Provider Adapter2.7%estimated ± 7.3 pp, medium confidence
SkillsBench official leaderboard (BenchFlow)56.0%estimated ± 6.1 pp, medium confidence
BinaryAudit43.9%measured
OTelBench15.9%measured
Vibe Code Bench14.3%measured
FACTS Multimodal52.7%estimated ± 4.0 pp, high confidence
ParseBench-273.6%estimated ± 5.0 pp, medium confidence
CharXiv81.4%measured
CharXiv w/o tools85.3%estimated ± 1.2 pp, high confidence
AI2D_TEST94.1%measured
C-Eval93.9%estimated ± 1.0 pp, low confidence
Claw-Eval51.8%estimated ± 4.5 pp, medium confidence
CountBench97.3%measured
DynaMath85.1%measured
ERQA70.5%measured
MathVision86.6%measured
MCP Atlas29.2%estimated ± 8.6 pp, medium confidence
MLVU (M-Avg)83.0%measured
MMLU-Redux97.9%estimated ± 0.8 pp, low confidence
MMMU-Pro81.0%measured
NL2Repo43.2%estimated ± 4.7 pp, high confidence
OmniDocBench 1.588.5%measured
PostTrainBench v118.1%measured
RealWorldQA83.3%measured
ScreenSpot Pro72.7%measured
SimpleVQA73.2%measured
SuperGPQA100.0%estimated ± 9.3 pp, low confidence
Toolathlon50.1%estimated ± 7.2 pp, medium confidence
V*88.0%measured
Video-MME (w/o subtitle)87.7%measured
Video-MME (with subtitle)88.4%measured
VideoMMMU87.6%measured
We-Math86.9%measured
AA LiveCodeBench91.7%measured
HealthBench Hard28.3%estimated ± 8.6 pp, low confidence
MedXpertQA (MM)76.6%estimated ± 2.4 pp, high confidence
React Native Evals61.9%estimated ± 3.0 pp, low confidence
ResearchClawBench18.0%estimated ± 2.2 pp, medium confidence
SWE-Rebench7.6%estimated ± 6.5 pp, low confidence
PostTrainBench15.7%estimated ± 1.2 pp, low confidence
BioMysteryBench (human-solvable)82.7%estimated ± 0.6 pp, low confidence
HealthBench (raw)48.4%estimated ± 6.0 pp, medium confidence
HealthBench (length-adjusted)53.5%estimated ± 3.4 pp, medium confidence
HealthBench Professional56.8%estimated ± 4.2 pp, medium confidence
HealthBench Professional (raw)59.1%estimated ± 4.5 pp, high confidence
HLE-Verified25.8%estimated ± 2.0 pp, low confidence
LABBench283.1%estimated ± 1.8 pp, low confidence
ScarfBench47.2%estimated ± 12.3 pp, low confidence
BlenderWorld (text-only)83.6%estimated ± 4.3 pp, medium confidence
Cookie-Bench90.3%estimated ± 4.8 pp, medium confidence
STEMBench2.6%estimated ± 0.2 pp, medium confidence
SimpleQA34.0%estimated ± 7.8 pp, low confidence
BabyVision w/ Python89.1%estimated ± 3.8 pp, low confidence
OpenHarmony Bench v1.053.1%estimated ± 3.2 pp, high confidence
BabyVision49.7%estimated ± 3.7 pp, high confidence
MMVU79.0%estimated ± 2.7 pp, medium confidence
PerceptionBench53.5%estimated ± 3.6 pp, low confidence
ZeroBench15.7%estimated ± 3.6 pp, low confidence
MRCRv284.8%estimated ± 11.3 pp, low confidence
CC-OCR79.0%measured
MMLongBench-Doc60.5%measured
RefCOCO (avg)84.1%measured
OCRBench V272.5%estimated ± 2.1 pp, medium confidence
BioBasicsBench33.9%estimated ± 2.8 pp, medium confidence
MMMU-Pro w/ Python82.9%estimated ± 2.0 pp, high confidence
ExploitBench Internal Port19.9%estimated ± 13.8 pp, low confidence
MathVision w/ Python93.2%estimated ± 0.5 pp, low confidence